Is a Second Pass Through the Sorter Worth It?

by ,

Sorting green coffee is time-consuming and until recently much too expensive to even test the idea of roasting beans of an identical hue. However, using sophisticated full-color optical sorters that eliminate defects in a single pass, STiR Tea & Coffee International asks: Is a second pass through the sorter to separate beans by color worth the effort?

By Dan Bolton and Jenny Neill

There are approximately 150 million green coffee beans in a shipping container – no two are identical, but many are similar in color.

Would cupping scores improve if roasters could sort and selectively blend the dark, medium, and lightly colored green coffee beans by hue?

Isolating the best from the rest is a challenge that begins with selective harvesting. The benefits of picking perfectly ripe cherries cannot be overstated (as few as .5% of green cherries in the lot negatively impact flavor). Quality controls continue through hulling, polishing, and mechanical grading by size. The final step is tedious and labor-intensive culling by hand, known as garbling, often in combination with optical sorters.

The Intercontinental Exchange, Inc. allows up to 23 defects per 350-gram sample. The Specialty Coffee Association of America (SCAA) Class 1 guidelines are far stricter: a bag of Class 1 beans has five secondary defects or fewer in a 350-gram sample. To reach that level of quality, producers need to rid the lot of many more flaws, including black, faded, sour, moldy, or crystallized beans. Nipped beans and those showing insect damage, immature beans, shrunken beans, and crushed beans must all be discarded to reach this highest class, one of five in the SCAA guidelines.

The reward for defect-free coffee is a substantially higher price over the approximately $2 per pound paid for commodity coffee. Defect-free coffee may look great but frequently scores well below the 80 out of 100 points necessary to command prices as high as $25 per pound. To get the highest price at auction the coffee must not only be clean, it has to display distinctive attributes in taste, aroma, body, acidity, and after taste. Roasters who perform a second sort by color may enhance that distinctiveness.

Beyond class 1

During a demonstration of the capabilities of the latest generation of optical sorters at Satake USA’s Houston headquarters, STiR wondered if there are any advantages in sorting for uniform characteristics beyond SCAA Class 1? The defect-free coffee remained a mix of colors. STiR asked: Would sorting by color make it possible for roasters to improve cupping scores?

“Distinctiveness (whether a coffee is in any way arresting or different in character) is a major influencer of scoring,” writes Ken Davids, editor of www.CoffeeReview.com. But he cautions that higher retail prices do not consistently reflect the price paid for green coffee.

“If the clarity of the cup were improved through your sorting procedure, and if the roast were executed well, then one might expect an improvement of a couple of (one to two maximum) points in a Coffee Review rating, given a similar improvement in average Q-cupper scores for the same coffee,” said Davids.

“However, roast is a factor, often a major factor, in adding or subtracting (cupping) points on Coffee Review. Furthermore, processes like hue-sorting may improve quality, but they may not create distinctiveness (whether or not a coffee is arresting or different in character), which is a major influencer of scoring. In my experience one cannot inject powerful and engaging character into a green coffee through improving its cleanliness and transparency, though you can let it be itself more clearly,” he said.

During the past three months STiR Tea & Coffee, in an experiment underwritten by Satake USA, Batdorf & Bronson Coffee Roasters, Nossa Familia Coffee, and Loring Smart Roast, has explored the potential benefits of starting with defect-free coffee followed by additional sorts based on hue.

The experiment began on June 27 with a 60-kilo bag of a specialty grade Nicaraguan Arabica.

The experiment

Once  graded  for  size  and  defect-free, a close examination showed some of the green beans in the bag were dark green in color, others were light green with the majority midway between. A control sample of the mixed beans was removed.

 Satake’s technicians then performed a second sort that isolated the dark beans and a final sort to separate the pale beans. The sorts were weighed and divided for shipment to Batdorf & Bronson and Nossa Familia in Portland for roasting and an evaluation by cuppers using the standard SCAA scale.

On receiving the coffee roast masters identically roasted these beans grouped by hue. The three samples were cupped and compared to the pre-sort control sample.

Scores from two independent panels of cuppers were compared to see if bean color makes a difference in the cup.

It did.

The results raised a number of questions:

Was the darker green color an indicator of quality? Did one color consistently receive higher scores? Are faded, lightly colored beans inferior?

Is the cupping differential significant enough for buyers to ask for a sort by hue? How much extra time and money would the additional sorts cost? Would roasting uniformly colored beans raise cupping scores enough to justify the cost?

The roasters

Uniformity is prized by roasters.

Constant air temperature and humidity are ideal. Uniform drum speed in the roaster, surface temperature, and a uniform first crack is ideally followed by uniform cooling.

Uniform bean size and density are critical. Outliers that are too small or too light end up charred. Beans too large or dense roast to a lighter color than desired.

“Roasters try to match a profile so that the line plotting temperature over time line up, the changes in color and the audible reaction from the first crack happen at the same time and at the same temperature every time, everything as close to mimicking the ideal roast as possible,”  said Rob Hoos, a roasting consultant and head roaster at Portland’s Nossa Familia Coffee. The roaster is owned by Augusto Carneiro whose family has grown coffee in the Sul de Minas and Mogiana region of Brazil since the 1890s.

Hoos used a 15-kilo Loring Smart Roast S15 Falcon roaster to precisely roast four of the eight samples from the sort.

Charge weight was 106 ounces (6.6 lbs.) for all samples. The charge temperature was 280-degrees and the beans were dropped into the cooling tray at 412-degrees, losing a uniform 14% weight after 11:49 minutes in the roaster. First crack was nearly identical for all four sorts at 9:44 minutes into the roast.

Hoos carefully mimicked the roasting curve for each sample, adjusting heat as required “so that the only difference we should be tasting in the coffee is correlated to its hue.” “Normally we don’t sort by color, this is something new,” he said. He observed that bean size varied slightly within the samples and the coffee emerged from the roaster as expected.

The profiles matched within an acceptable window of variation to ensure that any cupping differences were not due to the roast, he said.

Bob Benck, green bean buyer at Batdorf & Bronson, in Olympia, Wash., used a  Probat  BRZ-2  gas-fired  dual  sample

roaster. Benck aimed to have roast times and relative first cracks that approximate the roast results from their G-45 production machine. However, Benck cautioned that with their sample-sized equipment, an operator relies on sight and sound.

“Frankly, you should be able to do this with any coffee roasting equipment and get meaningful results as long as the roasts are done in identical fashion,” said Benck.

First crack for all four of the Batdorf & Bronson samples occurred after eight and half minutes and roasting stopped near the eleventh minute. Moisture readings were all within a .3% range.

Meanwhile in Portland, Hoos couldn’t wait to taste and compare. His notes show a marked improvement in scores for the dark and medium coffee.

“This was a decent coffee from the start (sometimes I am concerned that experiments don’t use decent quality coffees),” said Hoos. “If we were to do another set of roasts, I am sure we could slightly modulate the flavor but what we were doing here was looking beyond the roast (and trying to minimize its impact) and taste the difference in the green… specifically the impact of the hue sort,” he said.

“There was definitely a correlation between the hue sorting and the quality of the coffee when sample roasted and cupped. The fact that the deviation in score was close to or slightly over two points makes me believe that it isn’t cupper error. This is significant because buyers would normally be willing to pay more for an 86 as opposed to an 84, wouldn’t they?” he asked.

The next step was to see if seven other experienced cuppers would agree with his evaluation.

The cuppings

Three of the tasters were Q Graders licensed by the Coffee Quality Institute where they must pass a rigorous six-day course involving three-day examinations to earn their certification. There are 20 sections on coffee related subjects from identification to roasting and training in sensory skills and sensory triangulation. Q Graders are highly trained and calibrated in using SCAA cupping protocols. There are approximately 4,000 worldwide.

Tasters reported differences between the control mix and single hue samples but the cupping results overall were inconclusive. One group scored the lightly colored beans lowest among the samples, the other awarded the lightly colored beans the highest score. To learn more about cupping variables See Pg. 46, Descriptive Analysis Tools for Coffee, a discussion of how cuppers and researchers approach flavor calibration differently and why.

“In my opinion, the sample with the mixed hue tasted good,” said Midori Hartford, North America import manager at Sustainable Harvest. [It had] more depth and complexity in the cup.

“The original intent was to narrow the range of color so that the roaster could hone in on a specific profile therefore producing more reliable result. I know that if that sweet spot for a coffee gets really narrow through scrupulous sorting, then that target can be increasingly difficult to hit,” said Benck. “I am sure we are all up for that type of challenge though. It’s what keeps us coming back to the office everyday right?” he said.

Cupping scores for samples roasted in the Loring roaster were higher for the dark green sort (85.1) suggesting that the light green beans brought down the average. Hoos recommended blending the dark and medium colored beans and discarding the light hued beans. He suggested a light roast for the mix of medium and dark hue beans. The lighter beans are more suitable for dark roasting, he said.

Sustainable Harvest’s tasting notes: The Sustainable Harvest tasters found the blended hue “very slightly rubbery/pungent spice with allspice. The light hue was described as slightly tannic and very mildly herbal/hoppy. The medium hue roast was “very caramelized in fragrance and taste, almond brittle.” The dark is my favorite by far wrote Michael McIntyre, an independent Q Grader, “it was also the highest scoring mechanically… significantly sweeter.”

Batdorf’s cupping scores showed the greatest improvement in the light green hue which bested the pre-sort average by 3.25 points. The medium hue scored a point higher than the pre-sort blend. In cuppings, the light and medium green hues outscored the blended control. The point spread from the Batdorf & Bronson cuppers was too small to be significant, said Benck.

Batdorf & Bronson’s tasting notes: The pre-sort blend was described as toasty, malty, silky with a nutty finish, cracker, bakey, cocoa, and black pepper. Descriptors for the light hue included: sweet apricot, nutty finish, cherry, almond, orange peel with a mild body.” The medium hue brought these comments: “molasses, caramel, maple, almond, gritty, and dry. The darkest green was described as “caramel with lingering orange, floral, spicy” but tasters noted inconsistencies cup to cup.”

Recommendations

“I think that a roaster should still separate out the cup profile by roast hue, then find out which mix of hues and which ratio brings out the best of those specific beans to use in production,” said Hartford.

The additional cost for sorts is a modest $35 per hour. Neither roaster recommends the additional expense for coffee that does not already cup near the specialty threshold of 80.

Roasters replicating this experiment should consider evaluating the three hues independently instead of subjecting them to the identical roast. Doing so may further improve cupping scores. The  most practical  application is to treat the least desirable  hue as a defect. The beans would not go to waste and scores for the remaining hues would rise if the roaster succeeds at his task.

Successfully roasting a container sorted by hue could literally make the sum of its parts greater in value than the whole.

Post script

The fact that cuppers disagree on the individual scores is not as significant as the fact that cupping scores improved after sorting the coffee by color and that different roasters coaxed different characteristics out of the same colored samples.

“For me the biggest question is what causes the variation in color in the first place,” writes Benck. In reply grower Erwin Mierisch explains that “The slower the drying time the greener the bean. It all has to do with water activity. If we dry fast, the moisture in the nucleus of the bean is still very active. When the coffee rests and stabilizes that water migrates outward and raises the humidity and fades the color of the bean.”

Credits: Special thanks to cuppers Bob Benck, Brian Meyers, Arturo Villalobos, Aaron Shively at Batdorf & Bronson; Rob Hoos at Nossa Familia Roasters; Dane Loraas, and Midori Hartford at Sustainable Harvest; and independent Q Grader Michael McIntyre.